The conference has been quite a success so far, kicking off yesterday with a day of tutorials followed by a tapas-fueled welcome party in the center of Albecete. Today has been a full day of invited keynotes, contributed talks and posters on R applications, packages and practices. Here are just a few personal highlights from my notes:
Håvard Rue's keynote described the R-INLA package, which can quickly fit a subclass of Bayesian models (Latent Gaussian Models) without the need for MCMC.
Hugh Shahanan used Microsoft Azure's cloud service to automate the analysis a terabyte of microarray data in parallel with custom R scripting.
Constantino Garcia used time series analysis of the period between heartbeats (usign ECG data) to identify when patients were encountering bouts of apnea and other diseases.
Rob Hyndman described an elegant method of forecasting hierarchical time series, that provides a solution to the “bottom-up” vs “top-down” approach to forecasting a company's overall performance based on the forecast of individual departments and divisions. Chapter 9.4 of the free online book Forecasting: principles and practice describes the technique in detail.
There were many other excellent talks, and I'm looking forward to many more tomorrow. More later.